[llvm-commits] [llvm] r158226 - /llvm/trunk/lib/Transforms/Scalar/Reassociate.cpp

[Duncan Sands]

Hi Charles, I wasn't able to reproduce this :(  Maybe the following works:
run the hanging stage 2 command by hand, adding the following options:
   -emit-llvm -mllvm -disable-llvm-optzns
This should then generate bitcode in ClangAttrEmitter.o rather than object code.
Then see if running "opt -O2 -disable-output ClangAttrEmitter.o" hangs.  If so,
please send me ClangAttrEmitter.o.

Thanks, Duncan.

On 13/06/12 05:05, Charles Davis wrote:
> Sorry it took so long for me to get back to you.
>
> On Jun 11, 2012, at 1:12 AM, Duncan Sands wrote:
>
>> Hi Charles,
>>
>>> I don't know about you, but I'm still seeing this problem even with this fix (i.e. Clang hangs in Reassociate while compiling ClangAttrEmitter.cpp).
>>
>> I guess it's not exactly the same problem then.
>>
>>> Do you want me to revert (again) and file a bug?
>>
>> Maybe just explain how to reproduce it and I will try to fix it today.
> All I did was try to self-host Clang.
>
> Maybe I should give some specifics:
>
> - OS: Mac OS X 10.6.8
> - Stage 1 compiler: Clang
> - Stage 1 options: --enable-optimized --disable-assertions --enable-libcpp CXXFLAGS='-std=gnu++11'
> - Stage 2 compiler: Clang (Stage 1)
> - Stage 2 options: --enable-optimized --disable-assertions --enable-libcpp --disable-timestamps CXXFLAGS='-std=gnu++11'
>
> It usually hangs (though not always) when trying to build ClangAttrEmitter.cpp during stage 2. Whenever I profile the hanged Clang process, it's always hung in Reassociate.
>
> Chip
>